In short
Podcast Summary: The 404 Media Podcast - Episode: An Elon Musk-Funded PAC Is Running a Democrat Impersonation Campaign
Overview The episode discusses the implications of an advertising campaign funded by a super PAC associated with Elon Musk, which aims to impersonate Democrats and mislead registered Republicans about policies attributed to Kamala Harris. The discussion also touches on the use of AI in healthcare, particularly focusing on Northwell Health, New York's largest hospital system, and its integration of AI technologies.
Key Segments
- Elon Musk-Funded PAC and Political Manipulation
- Campaign Overview: The PAC, named Progress 2028, is designed to impersonate Democrats and present exaggerated progressive policies that scare registered Republican voters.
- Content and Messaging: The PAC's ads depict extreme policies like mandatory gun buybacks and open borders, which do not reflect Kamala Harris's actual positions as she has moderated her stance to appeal to a broader audience.
- Advertising Tactics:
- The campaign employs micro-targeting strategies across various platforms, making it appear as though Democrats have radical agendas.
- Text messages and Facebook ads are being used to convey messages that incite fear and anger among conservative voters.
- Social Media's Role in Political Campaigns
- Facebook's Handling of Political Ads: A discussion on how Facebook has shifted its approach to moderating political content, with concerns that misinformation may be proliferating without sufficient oversight.
- Comparison to Past Elections: Reflects on changes (or lack thereof) in social media moderation since the 2016 election, emphasizing the problematic nature of current advertising policies on platforms like Meta.
- AI Integration in Healthcare: Northwell Health Case Study
- Introduction to Northwell Health: The largest healthcare network in New York, employing over 85,000 individuals.
- AI Hub Presentation:
- Internal training revealed the hospital's use of a proprietary AI tool, AI Hub, enabling clinicians to utilize various large language models (LLMs).
- Concerns raised regarding the implications of using AI for clinical tasks, including summarizing patient data and potentially diagnosing medical conditions.
- Ethical Considerations:
- Questions about patient data privacy and oversight of AI suggestions were raised by medical professionals during the presentation.
- The potential for misuse and the risks associated with delegating clinical decision-making to AI systems were highlighted.
Key Takeaways
- Political Manipulation: The episode underscores the potential for manipulation in political campaigns through misinformation and impersonation tactics funded by influential figures like Elon Musk.
- AI in Medicine: While AI has the potential to improve healthcare efficiency, there are significant ethical and practical concerns regarding its implementation and the risk of over-reliance on automated systems in clinical settings.
- Call for Greater Oversight: Both discussions prompt a need for more robust oversight and transparency in political advertising and healthcare AI integration.
Conclusion This episode of The 404 Media Podcast effectively explores the intersections of technology, politics, and ethics, urging listeners to consider the implications of misinformation campaigns and the integration of AI in critical sectors like healthcare.
Listen to more episodes: For deeper insights and additional discussions, listeners are encouraged to subscribe to 404 Media at [404media.co](https://404media.co).
Written by AI. May contain mistakes. Listen to the episode to check what was said.
Transcript
Automatic transcript. May contain errors.0:00The very first thing we did when we were setting up 404 Media was incorporate our LLC and get compliant. When we were thinking about doing this, we actually delayed for a few weeks because I thought it was going to be super complicated. Then I found this week's sponsor, LegalZoom, which made setting up the business and getting it legally compliant a total breeze. 404 Media has been using LegalZoom since way before they became a sponsor. My favorite part of it is that it's got everything you need to launch, run, and protect your business all in one place. Anytime there's a question about the business, I can log in and find all our paperwork and compliance documents in one portal.
0:38Setting up your business properly and remaining compliant are things you want to get right from the get-go. But you don't have to strain your brain or wallet. LegalZoom saves you from wasting hours making sense of the legal stuff. At LegalZoom.com, you can take care of your business's legal needs in just a few clicks. And if you need some hands-on help, their network of experienced attorneys from around the country has your back. Over the last 20 years, LegalZoom helped start, run, and protect millions of businesses, including ours. LegalZoom. Now you're in business. Launch, run, and protect your business to make it official today at LegalZoom.com and use promo code 404media to get 10 % off any LegalZoom business formation product, excluding subscriptions and renewals.
1:25Expires 12-31-24. Get everything you need from setup to success at LegalZoom.com and use promo code 404media. LegalZoom.com, use promo code 404media. LegalZoom provides access to independent attorneys and self-service tools. LegalZoom is not a law firm and does not provide legal advice except where authorized through its subsidiary law firm LZ Legal Services, LLC.
1:56Hello and welcome to the 404 Media Podcast. I am Jason Kebler. We are Joe-less today, so I will be hosting this. I have with me Emmanuel. Hello. And Sam. Hi. and uh i guess the first housekeeping thing is we have socks available for sale uh on our merch store so if you've been looking for something to cover your feet uh to um you know stay uh warm in the fall and also modest in the fall you can check that out on our merch store you can find that at 404media.co they are modest they cover your ankles they do yeah sam is the model of the for the these and I think there's a couple photos of me as well and then a very threatening drain photo but we posted that on our site last week so you can go check that out this week we are going to be talking about the election and Facebook's role in it and also Elon Musk's role in it.
3:06And then we're going to talk in the second half of the show about how AI is invading hospitals, is being pushed on doctors. So that's what you're in for. I guess we'll start with my story. Elon Musk funded PAC Supercharges Progress 2028 Democrat impersonation ad campaign. this is very like complicated to explain i feel like that in a headline uh i wanted to use the term false flag i think emmanuel and sam both didn't like that term and i think that they were correct because a false flag is usually like conspiracy terminology for a specific irl event i also thought about calling it a honeypot or a psyop which maybe psyop is correct i would say It's a psyop for sure.
3:59Yeah, it's a psyop. So I guess I will explain what's going on here. So at the end of September, a website popped up called progress2028.com. And if you go to progress2028.com, it says two Americans ready for progress regarding how we build a sustainable engine of progressive change to carry us to 2028. and then you sort of scroll through and it is like it's not that long but it's basically like Kamala Harris's plan for her presidency which includes like quote empowering undocumented immigrants building our future mandatory gun buybacks sort of like open border vibes trans healthcare for all, including children, no parents allowed, and things like this.
5:03And I think if you're just like scrolling this page quickly, it sounds like sort of run-of-the-mill progressive ideas. But if you read it closer, it is sort of like an ultra-progressive like strategy that I would say is like a caricature of what conservatives are scared of, more or less. It's like a lot of it aligns very closely with the attack ads that have been running against Kamala Harris all over the country and especially in swing states. And I think that a lot of these policies are probably what, you know, very progressive of people would like to see. Like, I think mandatory gun buyback, very good idea.
5:56I would like to see that. But that is like, as many people know, Kamala Harris has like really moderated her opinions and sort of like run to the center in this election to try to get a lot of Republican votes. She's like campaigning with Liz Cheney, so on and so forth. It's like, she does not want mandatory gun buyback. She's like on every podcast saying that she has a Glock, more or less. And she's used it. She loves to say, yeah, she's like, and I use it. Yeah, well, so there you go. Yeah. She's also like tried to say that she's going to be tough on the border. And she has like, you know, basically moderated every opinion.
6:37And these opinions that are being promoted on Progress 2028 are not positions that she's really like ever held as far as I know. Most of them are, you know, I'm going to fully ban fracking and ban internal combustion engines for cars is like not a position that she has ever seriously held as far as I know. So in any case, like this, this website pops up at the end of September. It's called Progress 2028. It is clearly like an attempt to create the democratic version of Project 2025, which Democrats have been very successful in running on, which is, you know, this like thousand page document, very serious, serious in terms of like, a lot of thought has gone into it.
7:31put together by the Heritage Foundation, which is a conservative think tank, and have been able to pin this on Trump. And essentially what this seems to be is like a document was created that conservatives can point to and say, look, this is Kamala Harris's secret plan for 2028. And this pops up at the end of September. I got a tip on Signal from someone who told me about the article I wrote two weeks ago about this Elon Musk super PAC targeting Muslims in Michigan with messaging saying that Kamala Harris is pro-Israel. And then in Pennsylvania, Jewish people were getting ads saying that she is like pro-Palestine.
8:24So they're completely, you know, diametrically opposed messages that are being micro-targeted. And he said, hey, look, like, Progress 2028 exists, and it's the same PAC that's doing this. So it's the same Elon Musk-funded super PAC that is putting this together, and they are essentially targeting conservative voters with this ultra-progressive messaging that is, like, designed to inflame them and get them to be angry about Kamala Harris. Harris's secret plan. Does that make any sense? I feel like that took a very long time to explain, but it's like somewhat diabolical. It makes sense. But I feel like we also like, that's just the setup of like what the story is.
9:09So that's how like, how deep this goes. So yeah. Do you want to talk a little bit about the ads that this pack is running? Because I think that the copy on these ads was wild. If you want to get into what they are and maybe read some of them yeah so they um they set up this website and then they started like texting people um and that's like that's sort of when i learned about it then the website open secrets which it uh studies finance in political campaigns did a story tying progress 2028 to um the Elon Musk funded super PAC, which is called Building America's Future. And they were very tricky about how they did this.
9:55Like they registered to run Progress 2028 as a fictitious name under the PAC, which is insane that that is like a thing that you can do. You can basically like set up this dark money thing where someone funds one organization and then that organization funnels money to another organization. And then that organization sets up a subgroup under a fictitious name. And it's very difficult to track who is funding this, how they're doing it, who's behind it. So I know that's more setup, but basically in the last few days, they've started running a huge number of Facebook ads. And they're micro-targeting this stuff.
10:37We don't know how they're targeting it because Facebook's ad library really sucks. It doesn't tell you the parameters by which they're being targeted. But we can tell that they're micro-targeting it because they're running like 56 different versions of about a dozen different ads. So it's a total of 900 different ads that are being targeted to like 900 different groups. And here's one. It says, imagine a world where the American dream has no borders. and then it's like a bunch of people lined up at the border. They all look Latino to me in this picture. It's made to invoke the Mexico-U.S. border for sure.
11:23And then it says, Kamala has. See her vision for our future. And then the text of the ad says, we want to build an America where everyone, regardless of where they were born, has the opportunity to thrive. And Kamala Harris shares our vision. Drop a comment to thank Kamala. Another ad says, we're beyond thrilled to have Kamala on our side for the ban on fracking. Did you know she wasn't just a supporter of the Green New Deal? She was the very first 2020 presidential candidate to enthusiastically champion it. And then the picture there says, imagine a world without gas powered vehicles. Kamala has.
12:03There's a few ads about trans healthcare for children. There's a few ads about mandatory gun buybacks. And all of these have started running on Facebook in the last four days. I think the one with the gun dissolving is my favorite. It's like when Thanos snaps his fingers fading. It's also, I mean, all of these are so on the nose and the one with the gun is that way in that it is a pistol, right? It's not an AR-15. It's not, it's like the most extreme version of gun control. Perhaps it is a Glock. I don't know. I can identify the pistol. It's like the thing that you would like concealed carry. Like you can see like someone like getting like really pissed off.
12:54Like a conservative being like, you will not dissolve my gun. Everything is almost, I think as Jason says in the article, parody of democratic positions. Yeah. And I mean, the interesting thing also is that I have noticed that the Facebook ads are a little bit less extreme than the text messages. Like the text messages say, they use the same imagery, but a Republican registered voter in Pennsylvania sent me a bunch of texts that he got over the last few days from this group. And it uses the same dissolving gun, like one of them uses the same dissolving gun picture, but then it says, like, it basically says Kamala Harris supports a nationwide gun buyback program.
13:46And it says, like, mandatory gun buybacks in a few of the texts that I've seen. And it's having the intended effect. And we know that because there was a post on the gun politics subreddit that was like, vote like your rights depend on it. And it was a picture of a screenshot of that text message where it was talking about mandatory gun buybacks. And the person who got this text responded to that text message and said, quote, I have a nice warm bowl of cat shit you commies can share over the Canadian border where you belong. Fuck you from me in America. And then like all the comments are like, oh my God, I can't believe Kamala Harris is going to do that.
14:33And then over time, people realized that this was like an Elon Musk led psyop. But like the intended reaction is that conservatives are going to get messages like this and then they are going to be very mad and they're going to like share it with their friends and say like, look what Kamala Harris is going to do. Um, and I think the fact that the text messages are like a little bit more extreme is very concerning because we have almost no insight into text message campaigning. It's like there's been a lot of articles about text message spam during this cycle, but with online ads and with TV ads and with mailers, there's like a federal database in some cases of election mail.
15:22And then in other cases, like with Snapchat, there's a political ad library where you can see the targeting information, you can see what the ad was. I think that Meta's ad library is like really lacking for the reasons that I mentioned, but it does show what the ads actually are and you can see them. And I think that that's really important. whereas with a text message ad it's like you cannot see what people are getting unless people are sharing them online there's no national database kept by AT &T or Verizon that shares like here's here's what people are seeing and here's how they're being sent and it's really interesting because one of the phone numbers being used to send these messages to conservatives previously used run-of-the-mill Republican messaging from this same group.
16:20And we know it's the same group because they were running the same ads on Facebook about menthol cigarettes and how Kamala Harris is going to ban menthol cigarettes, which we can then see on Snapchat and Facebook. And then like a week later, they're pretending to be Kamala Harris, which is pretty sloppy, but it is a pretty good indication that this is... these are like Republican funded PACs sending this. This is not like Kamala Harris doing this or a Democratic PAC sending this. And if you want to read more about that, you can check out like Alex Howard, who runs a site called Civic Text, did a really good article about why there should be a national database of text messages for political advertising.
17:05I think that's like really important. If we still have a democracy in four years, check that out. but yeah, I don't know. Is there anything else on this before, before we move on? Like, what do you think? Uh, I forget how long ago it was, but maybe it was last year. We wrote something to the effect of, um, we're in the fuck it area of social media moderation. And I think one of the examples we gave is that following 2016 and people being in a real panic about how Facebook had this, I think we can probably say now, like overstated role in getting Trump elected then. People are really mad about Facebook and enabling all these ads from Russia and like fake users from Russia and events created on Facebook to fool people to go to fake protests and stuff like this.
18:05And after all that outrage, Facebook paraded all these election war rooms for various elections around the world saying they're doing all this stuff about misinformation. I don't think we're seeing a ton of that for this election cycle. And I'm wondering, I don't know, what is our general assessment of how Facebook is handling... I don't know, election-related misinformation, even if it's coming from like super PACs and coming from people who are stated funders and supporters of one of the candidates. Yeah, I mean, we haven't really talked about this on the podcast and I don't even think we've talked about it amongst ourselves much.
18:53I think, so Facebook and Meta more broadly has said it doesn't want to have anything to do with politics. and so it has said outright that it's like we are going to not recommend political stuff to people unless you're following it and you can like opt out of it more or less i don't even know the specifics on it at this point because adam masseri uh who is in charge of instagram has said so many different things on this but it's like there was a washington post article where posts with the word vote in them systematically were shown to fewer people than, you know, non-political posts on Instagram.
19:39And I think that you can extrapolate that that probably extends to Facebook. There's also been a lot of journalists on threads who are saying like, I was banned for posting the word Hitler, even though I was just doing like, you know, reporting and talking about Donald Trump. And so I think that that's very bad. I think that they're handling this extremely poorly. I think that then at the same time, there are millions and millions and millions and millions of dollars being spent on political ads on meta platforms. And presumably this is not like, they're not silencing the ads, You can still pay for reach on Facebook.
20:21And so Facebook and meta platforms more broadly are like places where you can advertise to your heart's content. But if you try to share news or politics, like there's a very good chance that it's either going to not be shared as widely or that it's going to be like actively suppressed.
20:43and yet I feel like those platforms are less important than ever is sort of my personal feeling and then like what Elon Musk is doing right now is so mind-boggling it is so hard to even like know where to begin with like what Elon Musk is doing right now in terms of actively campaigning with Donald Trump, turning X into, you know, like a far right platform, like actively sharing disinformation to millions and millions and millions of people about all sorts of things, offering to pay people in Pennsylvania for voting, like with this weird sweepstakes that may or may not be legal, you know, all the PAC stuff.
21:29And then at the same time, being like a really one of like having his companies propped up by government contracts and talking to Vladimir Putin in this free time. Like, I don't even think this is like all of the things. And so it's hard to say how much of an impact X will have. It's like Twitter has always been smaller than those other platforms. And yet it has become like this active disinformation space that I feel like the Facebook question is almost beside, not beside the point, but it's like, how do you disentangle that? It's like the information ecosystem is so fucked up right now. It's way worse than it was in 2016.
22:15And it wasn't even clear the impact that, you know, fake news had in the 2016 election. I just, I don't even know how to like begin to parse it more or less. Yeah, it seems, I'm not, I'm also skeptical of how much impact it has, but it is incredibly depressing. And it is messed up in the same way that all our non-consensual nudity reporting shows that Facebook is messed up. There's just so many ads, and Facebook is clearly not doing a good job of looking at them. I think if you keep pushing Facebook on this, these ads are not going to fly for impersonation reasons. I am somewhat confident they will eventually take action on this, by which point it will be too late.
23:09But that's just the Facebook portion of it. Everything else you said is just like, and I hate to be this guy because it's such a tired point by now. but it's just like i can't believe this is just allowed during like this very long election cycle just like yeah pay people to vote and like just buy fake ads pretending to be the other candidates and and so on it's just not the case in countries where there's like a legally mandated like two weeks of or two months of campaigning or something like this just just bummer vibes bummer vibes all around as always is around the election yeah i think the last thing i'll say is like there's so much money being poured into politics now.
23:52And it's like, obviously, obviously, like if you live in Pennsylvania or Wisconsin or Michigan, you're just getting absolutely hammered with this stuff. And it's crazy how much money is being spent trying to target like increasingly small numbers of voters. But I think to your point, it's like this election season goes on for so long and there's so much money that it's like campaigns and PACs can just throw shit at the wall and try everything. And this to me, like, I have no idea whether this progress 2028 thing is going to be successful in any way, shape or form. But it's like, it seems like some idea that was just like, someone had one day like in a terrible meeting and they're like, yeah, fuck it, let's try it.
24:41It does not seem like a carefully considered political strategy to me, but what do I know? Okay, we will leave that there for the first half of the show. When we come back, we will talk about AI in hospitals.
25:06Lumen is the world's first handheld metabolic coach. It's a device that measures your metabolism through your breath. And on the app, it lets you know if you're burning fat or carbs and gives you tailored guidance to improve your nutrition, workouts, sleep, and even stress management. I've got Illumin. And first of all, it's super simple to use and set up. Every morning, I breathe into it to learn whether my body is burning mostly fats or carbs. I've used that information to help me plan my meals and exercise based on how my body is working that day. I use it before a workout to learn if I need to fuel up and before or after exercise or eating to see how my body is working.
25:47Lumen has helped me understand my metabolism, showing me how eating certain foods at certain times impacts how I feel and how I burn carbs and fat. It's helping me work toward increasing my energy levels, feeling better during workouts, and even sleeping better. So if you want to take the next step in improving your health, Go to lumen.me slash 404media to get 15 % off your Lumen. That's L-U-M-E-N dot M-E slash 404media for 15 % off your purchase. Lumen.me slash 404media. Thanks, Lumen, for sponsoring this episode. This show is sponsored by BetterHelp. My dog Mango just turned 13 and I was thinking this weekend about how grateful I am for him.
26:34He keeps me grounded and helps me take my mind off things by getting me to go touch grass. This month is all about gratitude and along with our pets and our friends, there's someone else who doesn't get thanked enough. Ourselves. It's sometimes hard to remind ourselves that we're trying our best to make sense of everything. And in this crazy world, that isn't easy. When I'm back from playing fetch though, I like to try to process that crazy world and my reaction to it with therapy. Starting therapy has helped me understand what I can control and what I can't, taught me positive coping skills and how to set boundaries.
27:10It's also helped me deal with my anxieties, big and small. If you're thinking of starting therapy, give BetterHelp a try. It's entirely online, designed to be convenient, flexible, and suited to your schedule. Just fill out a brief questionnaire to get matched with a licensed therapist and switch therapists at any time with no additional charge. Let the gratitude flow with BetterHelp. Visit betterhelp.com slash 404media today to get 10 % off your first month. That's betterhelp.com slash 404media. For a gift that's always on time and lasts a lifetime, you can't do better than Masterclass. With Masterclass, your loved ones can learn from the best to become their best.
27:57Masterclass is the only streaming platform where you can learn and grow with over 200 of the world's best. That's why Wirecutter calls it, quote, an invaluable gift. Gift unlimited learning. Learn from any masterclass instructor anywhere. On a smartphone, computer, smart TV, or even in audio mode. Help your loved one confidently navigate the media with the influential intellectual Noam Chomsky. Use science to solve problems with Bill Nye. Or learn from the past with Pulitzer Prize winning historian, Doris Kearns Goodwin. I'm going to check out a class by Amy Poehler about how to use improv skills in my everyday life.
28:35The classes really make a difference. 88 % of members feel that Masterclass made a positive impact on their lives. Plus, there's no risk. Every new membership comes with a 30-day money-back guarantee. Give your loved ones a year of learning with Masterclass. Masterclass always has great offers during the holidays, sometimes up to as much as 50 % off. Head over to masterclass.com slash 404pod for the current offer. That's up to 50 % off at masterclass.com slash 404pod. masterclass.com slash 404pod.
29:19Okay, and we're back. We're talking about a big scoop from Sam. leaked training shows how doctors in New York's biggest hospital system are using AI. Sam, the story is about a hospital system called Northwell Health. Who are they? Yeah, Northwell is one of... Actually, it is the biggest hospital network, healthcare network in New York. It has over 85 ,000 employees. If you live in New York, you probably have touched Northwell's network in some capacity. Yeah, so just to kind of give you the background of this story, a couple months ago, I got a leaked recording of this hour-long demonstration slash presentation from internal Northwell Health leadership.
30:16And this presentation was just like a training, kind of, or like a demo of this new tool that's like fancy new AI tool that they had called AI Hub. And they were encouraging staff to use it. They're kind of like trying to get people onboarded to use it. They were like, you know, we have a couple thousand people using it now and it's spreading word of mouth. They were like very excited about getting this thing popular within the network, I guess. And then I think that, I mean, I don't want to laugh at this person's title, but the title of the person giving the presentation was Senior Vice President and Chief of Digital Innovation and Transformation.
Read the full transcript
30:59To give you the kind of idea of what we're dealing with here. We're doing some innovation here. That's a Vice Media-esque employee title. Yeah, Senior Innovation Officer vibes. But yeah, so the innovation here is that Northwell built this tool internally that is basically like a closed system of LLM. So it's, you know, you have like this one dashboard that you can choose from a bunch of different large language models. I think the list is like 14 different models, but you have like Gemini, Claude, MedLM, which is like specifically for healthcare workers. uh cody which does coding um chat gpt of course and a bunch more so the pitch from leadership to clinicians and the people on this call were um doctors radiologists staff it's you know it's we're they're pitching people who actually have like interaction with patients in this particular presentation, um, is that you can use this tool to, I mean, I want to say they they're saying that you can use this tool to make their lives easier, but it's like, I can't imagine this making anyone's lives easier, which we can get into.
32:18Um, but yeah, I can just kind of run through like some of the use cases that they suggested for the friendly AI hub. Um, just to go through kind of quickly, they were suggesting you optimize and summarize things like meeting minutes, articles, emails, clinical charts. So we're talking about like patient data, protected patient data that includes like, you know, personally identifying information. They suggested you could use it for like parsing billing code or writing job descriptions or reading resumes, comparing resumes, translation, education materials, things like that. And also to kind of extract information out of documents.
33:07And the big kind of red flag here is that they're suggesting that you can, and they're encouraging and kind of reassuring these clinicians that it's safe to put patient data into this network, this chat GPT for doctors that they've built because it's not touching the open internet. It's just like, it's in the walls of the network is what they're describing it as. Yeah. The interesting thing to me about this story is that, and we can talk more about this in a second, but the presentation starts and they're like, oh, this can help you professionalize your emails. Like you put in, you type something, like maybe English is your second language or something, and you want to make it sound more professional.
34:01Like you can do that. Then they're like, you can also use this maybe to do some translation work or something like that. And then they sort of also say that you can use this to summarize clinical charts, which is like patient data. And then the person who was running this training or who was doing a lot of the training was scrolling through sort of like other potential use cases. And one of them was like, diagnose pancreas cancer. And then another one said parse HL7. And HL7 is like a data protocol, like a file standard for electronic health records. And I guess the thing that stood out to me was that they're pitching what I would consider to be pretty innocuous use cases, but then also teasing way more concerning and advanced things.
35:01So what can you say about that, Sam? Because it seems like this isn't being widely used at the moment to diagnose patients, but it's definitely like that's where the head is at is sort of my thought. Yeah, for sure. It's definitely, I mean, they, um, yeah, again, it's like, it's hard to really tell to what capacity they're using this in hospitals. But I think first of all, Northwell is one of like many different hospital systems that are using AI at this point. Um, so this is kind of an inside look at how this is being pitched to doctors who might be skeptical of how this is going to go. I thought it was interesting because one of the, like you said, one of the pitches was detecting pancreas cancer, pancreatic cancer.
35:53And a radiologist at that point in the demonstration kind of did the hand-raise thing and was like, is there any kind of like oversight, medical or ethical oversight of the publication of tasks because you can kind of publish out to the wider system what you're working on in case it's helpful to anyone else and when they're talking about like detecting cancer i'm sure every radiologist and most of the doctors and probably all the staff are like what like what are we doing here so i thought it was interesting that that was who raised that question was like okay what's the oversight what kind of ethical questions are there surrounding this um and you know that that radiologist went on to say, you know, I realize this is probably just for play right now, but at some point people are going to start to trust this to do medical decision-making, which is like, they're already seeing a couple of steps ahead and doctors are suspicious too, just like patients would be.
36:49And then at that point, the, one of the presenters answered the question and said, well, that's still being developed. And then for any kind of task that is going to have a clinical undertone or clinical adjacent. We're going to have a group of people overseeing it to make sure that it's not bullshit, basically. And then the question is, well, who are those people? Who is getting to see my chart or my scans or whatever it is that's not my doctor, not my medical team that I've authorized to do that? Is it just this innovation officer? Is it her staff um is it you know people that they've hired outsource um yeah it's and it's it's tough too because it's like there are promising studies for detecting cancer with ai it's been something that's in the news quite a bit recently that you know there are these like promising results in detecting cancer using ai models um especially with breast cancer which is like so important to detect early there's some promising uses there so it's not like ai is useless and evil in all capacities in health care but this is an example of a hospital kind of pushing every little use of ai no matter what just in a blanket way toward their doctors and saying use this and we'll figure out later, whether it's helpful or good or making shit up.
38:28I think it's just like, it's a wild kind of look into that world. Yeah. I mean, that's another thing that stood out to me where they're like, we've created these specific tasks that you can use, which is translating or writing emails or whatever. But then you can also create your own tasks. And there's kind of this... I don't know if they outright say it or not, but what I took away is like, have fun, innovate, try whatever. And there's 58 ,000 people who work at this, you know, in this healthcare system. It's like, what is the oversight here, I guess, is sort of my concern. Like, it doesn't seem like the hospital itself is going to turn on, you know, use AI to diagnose your patients tomorrow.
39:22But there's nothing stopping any doctor from doing that right now, from what I can tell, because it's available to everyone. It's live now. And it has this Google LLM, like the MedLM is a health specific LLM. So, I don't know, is it just like, go hog wild? I don't know. That was the playful aspect of this was my observation as well, which is, you would imagine everything at the hospital would be very methodical and intentional because it's life and death situations. And you'd imagine if they have, if they want to start trialing an llm then they would pick one they will be like we tested it for hundreds and hundreds of hours and this is the correct llm to use for this task and they're like we just like we got licenses for all 14 and like feel free to mess around and see which one you like which seems very weird and like i understand that um like some of the tasks are administrative but that stuff is really important in medicine also you know it's like whether you're translating something or sending emails like i've been at the hospital a lot lately for medical reasons and like they actually work on that stuff really hard like what is the best follow-up email to send to a patient in order to ensure compliance with like you know taking your medication after you leave the hospital and stuff like this and to just like mess around and like try to have Claude 3 do it or Gemini 1.5 or Chad GPT.
41:00It just seems very... I don't know if I'd call it sloppy, but experimental in a way that you wouldn't imagine a hospital would be broadly. One of the use cases where they're like, definitely do this because it's going to save you a lot of time is they're like, if you're on an email thread with 20 replies, just have it summarize the email thread into three bullet points for you it's like what like what uh there might be some important stuff in there you might need to know exactly what it is and then they're also like you can have it just like summarize meetings for you as well into like three bullet points so like didn't pay attention during this hour long three hour long meeting like here's three bullet points yeah i can see that being bad like that doesn't everybody everyone who listens to this probably knows, but it's like that probably works a lot of the time, but sometimes it doesn't.
41:52And you just don't want to be the case where it's like, oh, the AI summation tool forgot that Emmanuel needs to take his insulin or he will die or something like that, you know? Yeah. And then one of the other examples they had was Sam mentioned, they parsed this like 36 page document about how to interact with patients who are like remote, remotely monitoring patients. and one of the examples was like, tell me what this says about billing. And it says like, here's some bullet points. I can easily imagine a doctor just like going through that or whoever makes the bill, like doing that and then automating like, oh yeah, like here's what the AI told me we should charge this person.
42:35And it's like, I don't know, that doesn't seem good. That easily could lead to overcharging someone. And how often is a patient informed? And by informed, I mean, like, how often is a patient able to parse what a hospital should have charged them? It's like you just, you're at their, you have no power in that situation. You pay whatever they say, or you go into, like, lifelong medical debt, like, more or less. Yeah, I mean, the idea here, I think, the pitch, basically, is that, and this is not just Northwell, this is all hospitals are using AI. and AI in general being just kind of blanket pushed into these industries is it's going to be like some great time-saving tool.
43:21But this is something that like we, I think we've all talked about this in various times in the podcast and also the experts that I talked to for the story brought up that like burnout is extremely real with doctors. Doctors are very tired, very stressed out, and they already have so much workload and a ton of it is administrative. like they shouldn't be spending 80 % of their day staring at paperwork or doing billing bullshit or things like that they should be like interacting with patients and getting more context from like their actual visits but they can't because they're overloaded and somehow ai is going to come in and magically take away that burden but how i can kind of see this playing out is you have stressed out tired doctors just dumping like information into this new tool that their healthcare network has told them is accurate and good and they should use, ripping whatever came out of it out as accurate and good, even though it might not be.
44:19And in a lot of cases, it hasn't been for other LLMs and other systems. And then running with that. And then you have actual harms being done. You either have people doing administrative mistakes or, in the worst case, like diagnoses that are wrong or like, you know, care practices that are the wrong things to suggest that they're just like, whip it out into an email, whatever, send it. And it's like, that would be a really, that would be a really bad doctor. That would be malpractice. But like, that's kind of what they're suggesting to do is like, oh, you're crunched on time. Just let the AI do it.
44:57It's like, that's not good enough in so many cases. so yeah and like it's i mean using an internal system is definitely better than plopping stuff into chat gpt that's like hipaa protected information like no one wants doctors doing that so this is kind of the alternative that they're posing but it's like they're using it as a testing round on the doctors on the clinicians and then by association patients and patients probably don't even know that this is happening that their doctors are using this LLM tool to put their data into and then suggest, you know, whatever the standards of care are. So yeah, I don't know.
45:38It's very, it creeps me out. I don't like it. Yeah. I don't even know what to say about the patient privacy aspect of it because I do understand the argument that like this isn't connecting to the broader internet. So, you know, no problem here. It's not going to leak patient health records to the broader internet. But this is a massive, massive hospital. And so I don't know, is it going to leak into other patients' records? Is it going to further train the LLMs within the hospital? It's like, that's not super clear. I don't know. And I could see that being a problem. It's like, New York is a very large state yeah and big cyber attacks happen you know it's like this is like a thing that does happen is that internal systems get exposed to massive hacks so i don't know it's like it's i don't i don't think that's a whole lot better either i found the 85 000 employee number mind-boggling personally it's like that's a lot of people that is like on a scale of like how many people worked at Facebook a few years ago.
46:54And this is just one hospital network in New York. So it's like, it's a pretty big network. Yeah. If you're walking on the street in New York, you might bump into a Northwell employee more often than not. It's like in New York City. It's kind of how that plays out. But yeah, it's huge. And it's, you know, it's anyone with an email address that works at Northwell can access this tool. So it's not just doctors. Anything else on this? I don't know. I think we covered it. I mean, just don't get sick and don't pay attention to politics, I think is what we've established so far in this podcast. Yeah, yeah.
47:33Okay. If you are listening to the free version of this feed, we will end it there and play us out. if you are listening to the subscribers only version or if you want to become a subscriber we are going to talk about sketchfab and epic games's takeover of it in the bonus segment of this podcast sketchfab is a 3d model database that has lots of very culturally important models in it as well as like lots of game stuff and just 3D art in general. And it's now owned by one of the biggest video game companies in the world. That's Emanuel's story. We'll be talking about it after the break. If you are not a subscriber, you can become a subscriber at 404media.co and get access to that as well as bonus segment on every single one of our podcasts and some other bonus episodes that are for subscribers only.
48:36So go check that out. Let me do the outro now. Let me summon my Joseph Cox memory here. Thank you for listening to the 404 Media Podcast, a production of 404media.co. We publish every week. If you are a free listener, you can get access to bonus content and more at 404media.co. We are a journalist funded and run publication. No billionaires have supported this production or our website. This is made in production in cooperation with Kaleidoscope Studios. Did I miss anything? Nailed it. Really didn't nail it, but thank you for listening.
From the publisher
An Elon Musk-funded super PAC has expanded an advertising campaign in which it is impersonating Democrats and targeting registered Republicans with policies unpopular with conservatives they say Kamala Harris will pass if she wins the election. Plus, Sam takes us inside Northwell Health, New York's largest hospital system, which is pushing AI on its doctors.
Articles discussed in this podcast:
https://www.404media.co/elon-musk-funded-pac-supercharges-progress-2028-democrat-impersonation-ad-campaign/
https://www.404media.co/northwell-health-ai-hub-tool-chatgpt-doctors/
https://www.404media.co/scientists-and-archivists-worry-epic-games-control-of-the-3d-model-market-will-destroy-cultural-heritage/
Learn more about your ad choices. Visit megaphone.fm/adchoices
